On the success of a swindle: pollination by deception in orchids
- First Online:
- Cite this article as:
- Schiestl, F.P. Naturwissenschaften (2005) 92: 255. doi:10.1007/s00114-005-0636-y
- 1.4k Views
A standing enigma in pollination ecology is the evolution of pollinator attraction without offering reward in about one third of all orchid species. Here I review concepts of pollination by deception, and in particular recent findings in the pollination syndromes of food deception and sexual deception in orchids. Deceptive orchids mimic floral signals of rewarding plants (food deception) or mating signals of receptive females (sexual deception) to attract pollen vectors. In some food deceptive orchids, similarities in the spectral reflectance visible to the pollinator in a model plant and its mimic, and increased reproductive success of the mimic in the presence of the model have been demonstrated. Other species do not mimic specific model plants but attract pollinators with general attractive floral signals. In sexually deceptive orchids, floral odor is the key trait for pollinator attraction, and behaviorally active compounds in the orchids are identical to the sex pheromone of the pollinator species. Deceptive orchids often show high variability in floral signals, which may be maintained by negative frequency-dependent selection, since pollinators can learn and subsequently avoid common deceptive morphs more quickly than rare ones. The evolution of obligate deception in orchids seems paradoxical in the light of the typically lower fruit set than in rewarding species. Pollination by deception, however, can reduce self-pollination and encourage pollen flow over longer distances, thus promoting outbreeding. Although some food deceptive orchids are isolated through postzygotic reproductive barriers, sexually deceptive orchids lack post-mating barriers and species isolation is achieved via specific pollinator attraction. Recent population genetic and phylogenetic investigations suggest gene-flow within subgeneric clades, but pollinator-mediated selection may maintain species-specific floral traits.
Although pollination by deception has attracted great interest from ecologists and evolutionary biologists, its spectacularly diverse evolution among orchids still poses many unsolved questions. Here I review concepts of pollination by deception and recent, significant advances in our understanding of proximate and ultimate causes in this fascinating topic of ecology and evolution. I will limit this review to the pollination syndromes of food deception and sexual deception in orchids, because most research has focused on these two systems.
Floral mimicry: how to cheat a pollinator
Although generalized mimicry is assumed to be common among food deceptive plants (Dafni 1984), Batesian mimicry has been demonstrated only rarely among food deceptive species, but is well established in sexually deceptive orchids. Evidence for Batesian mimicry requires a number of assumptions to be met. Obviously model and mimic must be sympatric and co-blooming, and the mimic must occur at lower frequency than the model, since operators may eventually learn and avoid the mimic. The mimicry must be advantageous for the mimic, i.e. the reproductive success of the mimic must be higher when the model is present (Wiens 1978; Nilsson 1983; Dafni 1984; Ackerman 1986; Johnson 1994; Roy and Widmer 1999). This assumption is, however, difficult to discriminate from the “magnet species effect,” and thus studies should control for co-blooming, rewarding non-model plants. A crucial part of a Batesian mimicry is that the operator (pollinator) initially does not discriminate between the model and mimic (Wiens 1978), and hence the floral signals of model and mimic must be similar for the pollinators senses. Modern analytical and physiological tools have enabled the testing of this latter assumption, which has led to amazing insights into the mechanisms of floral mimicry.
Pollinators and floral signals
Food deceptive orchids are pollinated by a large array of different insects. Bees are the most important pollinators of many species, but butterflies, moths, flies, and beetles are also exploited by some orchids (e.g. Nilsson 1979; Nilsson 1980; Nilsson 1983; Nilsson 1984; Petterson and Nilsson 1993; Johnson and Steiner 1997; Steiner 1998; Roubik 2000; Johnson 2000). Visual signals are thought to be of primary importance for pollinator attraction in food deceptive systems. Behavioral experiments and field observations have shown that bees prefer, after being rewarded, flowers of similar color, thus indicating selection for corolla color similarity in mimics and potential model species (Gumbert and Kunze 2001; Gigord et al. 2002; Johnson et al. 2003a). Such selection may, however, be mitigated by the temporal and spatial dynamics of co-blooming plant communities, explaining the lack of a specific model species in generalized mimicry systems. A few examples of Batesian mimicry, however, are well documented, such as Cephalanthera rubra, Disa ferruginea, and D. pulchra and their respective models, where similarities between model and mimic in color reflectance spectra visible to the pollinator have been reported (Nilsson 1983; Johnson 1994; Johnson 2000). In a more elaborate approach, Galizia et al. (2005) mapped visual signals on the perceptual space of honey bees and confirmed similarities in Anacamptis israelitica and its suspected model Bellevallia flexuosa.
Many species of food deceptive orchids emit floral odor (Nilsson 1979; Nilsson 1983; Nilsson 1984; Bergström et al. 1992; Moya and Ackerman 1993; Ackerman et al. 1997; Barkman et al. 1997; Salzmann and Schiestl unpublished), but little is known about similarities of mimics and potential model species. In comparing floral odor of Anacamptis israelitica and B. flexuosa and its perception in the honeybee brain, Galizia et al. (2005) found no evident similarity in this floral trait. In behavioral experiments with bumble bees foraging on artificially scented flower dummies, Kunze and Gumbert (2001) showed that discrimination of model and mimic by bumble bees is greater if scents can be used to identify the mimic, however, discrimination is poorest if the floral odor of model and mimic is identical. Therefore, mimics should be selected to produce floral odor identical to their models or avoid odor production altogether. These experiments provide intriguing hypotheses that should be tested in future studies to better understand how odor influences floral mimicry.
Pollinator behavior and selection on plants
Pollinator behavior influences reproductive success and gene flow within and among plant populations and may thus impose selection on plants that are pollinator-limited in their reproductive success, a condition frequently found among deceptive orchids (Tremblay et al. 2005). Pollinator-imposed selection on floral traits is generally stronger when pollination is specific, as selection will be influenced by the more uniform behavioral patterns of one or few pollinator species. Specificity in pollination is common among deceptive orchids (Nilsson 1992; Tremblay 1992), and especially pronounced among sexually deceptive species (Paulus and Gack 1990; Bower 1996). An important aspect of pollinator behavior is the usually high capacity of flower visitors to associate deception with certain floral signals and thus avoid non-rewarding flowers. This is prevalent even in sexually deceptive flowers, were floral signals release innate behavioral patterns (Ayasse et al. 2000). Besides pollinator behavior, pollinator morphology also imposes selection on flowers and has, for example, been shown to drive the evolution of spur length (Nilsson 1988; Johnson and Steiner 1997), but here I will focus on behavioral patterns of pollinators.
Pollinator learning and negative frequency-dependent selection
Deceptive orchids often show a remarkable degree of variation in floral traits such as color (Nilsson 1980; Petterson and Nilsson 1993; Johnson 1994; Gigord et al. 2001) and scent (Ackerman 1986; Ackerman et al. 1997; Moya and Ackerman 1993; Schiestl et al. 1997; Barkman et al. 1997; Ayasse et al. 2000; Salzmann and Schiestl unpublished). A possible explanation for the maintenance of these polymorphisms is negative frequency-dependent selection. This type of selection is a likely outcome of the learning and avoidance of individual flowers or flower morphs by the pollinators, since common morphs will be avoided more quickly than rare ones and hence face fitness disadvantages (Heinrich 1975; Ackerman et al. 1997; Smithson and Macnair 1996; Gigord et al. 2001). In many floral mimicry systems, pollinators are able to learn and subsequently avoid the mimic, although they might initially not discriminate between model and mimic (Dafni 1984; Peakall 1990; Paulus and Gack 1990; Ayasse et al. 2000). An example of switching between different color morphs after an unrewarding visit are bumble bees that pollinate Dactylorhiza sambucina, a food deceptive orchid occurring in purple and yellow flower morphs (Gigord et al. 2001). This behavior leads to overvisitation of the rarer morph and may explain the maintenance of the two forms in stable frequency within a population. In a survey of natural populations in the same species, Pellegrino et al. (2005) found, however, no evidence for negative frequency-dependent selection, but a positive correlation between the frequency of yellow morphs and the reproductive success of both morphs. In an experimental approach with painted flowers, Aragon and Ackerman (2004) also did not detect negative frequency-dependent selection on flower color in the polymorphic Psychilis monensis. Besides flower color, pollinator-imposed selection may act on floral odor, a trait that is generally learned faster in bee-pollinators than color (Bogdany 1978; Menzel 1985). In Tolumnia variegata, however, negative frequency-dependent selection seems not to explain odor polymorphism, but in this case odor variation was only assessed as overall presence/absence of scent and not at the individual-compound level (Ackerman et al. 1997). Individual compounds in the floral odor of the deceptive Anacamptis morio vary significantly more compared to the rewarding A. coriophora, possibly as an outcome of negative frequency-dependent selection in the deceptive species versus stabilizing selection in the food rewarding species (Salzmann and Schiestl unpublished).
Potential for negative frequency-dependent selection on floral odor has also been shown in the sexually deceptive Ophrys sphegodes, where pollinators learn and subsequently avoid the odor bouquet of individual flowers (Ayasse et al. 2000). Although pollinator behavior often predicts negative frequency-dependent selection, its potential impact on complex traits such as scent is difficult to assess, because the behavioral significance of individual compounds is often unknown.
Pollinator preferences and selection
Among the floral traits potentially influenced by pollinator’s preferences, flower height and size have received most attention. Some studies on food deceptive orchids report a positive correlation between flower/inflorescence size and pollinaria removal and/or fruit set (Schemske 1980; Fritz and Nilsson 1996) whereas in others, flower size has little impact on fruit set (Sabat and Ackerman 1996; Pellegrino et al. 2005) or selection for smaller flowers is found (O’Connell and Johnston 1998). Floral height correlates positively with reproductive success in Cypripedium acaule (O’Connell and Johnston 1998). Pollinator-imposed selection may generally vary with abundances of different pollinator species over years, and different trade-offs and constraints may influence evolution of floral traits in the plants (Schemske 1980; O’Connell and Johnston 1998).
Flowers of the sexually deceptive Chiloglottis trapeziformis are significantly larger than females of the pollinator species and produce up to 100 times more of the attractive odor compound (Schiestl 2004). The pollinators prefer larger females and larger amount of scent, which may explain the evolution of such exaggerated floral traits (Fig. 2). In C. trapeziformis and C. valida the pollinators show a preference for dummy females presented at different heights matching the height differences in the respective orchid species (Schiestl and Peakall in press). The pollinator of C. trilabra also selects for floral height, but the natural mean height of plants is lower than the experimentally determined optimum for visitation, indicating pollinator-mediated selection cannot explain the precise floral height of this species (Peakall and Handel 1993).
Do orchids impose selection on their pollinators?
There is good evidence that pollinator behavior imposes selection on plants, but do plants also select for pollinator traits? In general, evolutionary change in orchid–pollinator interaction is best viewed as one-sided, i.e. plants responding to pollinator-imposed selection but not vice versa (Dafni 1984; Nilsson 1992; Wasserthal 1997). The evolutionary direction of a mimicry system is, however, determined by the fitness impacts the mimic may have on a model or operator species (Stowe 1988). In the Australian sexually deceptive Chiloglottis trapeziformis, pollinators are unable to discriminate between the odor of orchid flowers and the female sex pheromone (Wong and Schiestl 2002), and avoid areas where the unrewarding orchid flowers occur (Peakall 1990). Under certain circumstances, such as when populations of wasps and orchids largely overlap, this could lead to an interruption of the normal mating behavior of the pollinator insects. This will have fitness costs to the model, i.e. if the wingless females of the pollinator species are unable to obtain a mate when calling from within a patch of orchids (Wong and Schiestl 2002). The model may therefore be selected to avoid orchids, and field experiments suggest that females can respond by walking out of an orchid patch, which will resume their attractiveness for males (Wong et al. 2004). Walking a few meters after unsuccessful calling may have evolved specifically in species exploited by orchids, however, experimental tests are lacking.
Evolution of deception in orchids
Despite the strong interest in floral mimicry among evolutionary biologists, its spectacular evolution in orchids still poses a number of unsolved questions. It seems paradoxical that deceptive orchids often show considerably lower fruit set than rewarding species (Neiland and Wilcock 1998; Tremblay et al. 2005), e.g. only 2% in Cypripedium during a 10 year survey (Gill 1989), yet deception has evolved in a large number of species, many times independently within the family (Dafni and Bernhardt 1990; Cameron et al. 1999; Cozzolino et al. 2001; Kores et al. 2001; Bateman et al. 2003). Although sexual deception is clearly a derived trait among the orchids (Kores et al. 2001; Bateman et al. 2003), food deception, or at least nectarless flowers, may actually be an ancestral trait in the orchids, suggested by the nectarless flower of primitive orchid groups such as Apostasia and Neuwiedia, that are pollen-rewarding (Kocyan and Endress 2001). In certain genera such as Disa and Anacamptis, nectar reward has clearly evolved secondarily from food deception (Johnson et al. 1998; Cozzolino et al. 2001).
A number of hypotheses have been proposed to account for the evolution of deception, i.e. costs of nectar, low density, pollinia, pollinia loss, better isolation, and pollinia removal, described in Bergström (1978), Dressler (1981), Dafni (1987), Nilsson (1992), and Smithson and Gigord (2002), that are little tested or did not gain empirical support. One hypothesis that has been supported by recent empirical investigations is the outcrossing hypothesis, proposing that deception reduces geitonogamous pollination (pollen transfer between flowers of an inflorescence) and promotes longer distance pollen flow, because pollinators will visit fewer flowers on an inflorescence, and fly greater distances after being deceived (Dressler 1981). Geitonogamy may reduce female fitness if inbred seeds develop slower or are aborted in early stages of development due to inbreeding depression (Peakall and Beattie 1996; Johnson et al. 2004). Geitonogamy may also impact on male function, since less pollen will be available for export. Many orchids transfer their pollen in pollinia, making fertilization highly efficient, and thus the impacts of geitonogamy more severe. For perennial plants that produce thousands of minute seeds per capsule, like many orchids, the production of fewer, but high-quality fruits, and the maximization of pollen export may increase live-time reproductive success (Peakall and Beattie 1996). Field studies confirm that pollinators, when visiting deceptive orchids, usually probe a few flowers and then depart quickly (Nilsson 1980; Nilsson 1984; Peakall and Beattie 1996; Ayasse et al. 2000; Johnson 2000). In flowers of Anacamptis morio supplemented with nectar, bees probe more and stay longer on flowers, take up more pollinia and cause more self-pollination through geitonogamy (Johnson and Nilsson 1999; Smithson 2002; Johnson et al. 2004). Modeling pollen export in A. morio suggest that deceptive flowers export more pollen compared to rewarding ones, when pollinators are abundant and most pollinaria are removed from the flowers (Johnson et al. 2004). In the sexually deceptive Caladenia tentaculata, pollinators tend to leave a patch of orchids after pseudocopulation and 87% of pollinations are outcrossing (Peakall and Beattie 1996). Mean distance of pollen flow, detected by tracking stained pollinia, is 15.2 m, which is significantly greater than the mean nearest neighbor distance, and maximum distance of pollen flow is 58 m. In the sexually deceptive Drakea glyptodon, pollinators fly an impressive mean distance of 32 m (maximum 132 m) after pseudocopulation at experimentally presented flowers (Peakall 1990). Males of the solitary bee Colletes cunicularius, which pollinates four species of Ophrys, fly an average distance of 5 m during their patrolling flights, with maximum of 50 m (Peakall and Schiestl 2004). Thus, many studies, especially of sexually deceptive systems, indicate that deception promotes outcrossing, and the low-fruit set in these pollination systems should not be viewed as an indication of a sub-optimal life history (Gill 1989), as it probably is the outcome of a well-balanced trade-off. However, little understood examples are the clonal Leporella fimbriata, that is pollinated by pseudocopopulating male ants, and 88% of vector flight are within bounds of clones, and consequently high rates of selfing are expected (Peakall and James 1989). Thus, more data are required to conclusively identify selection driving the evolution of deception.
An alternative view, applying only to the evolution of sexual deception, is its opening of new pollination-niches. Sexually deceptive orchids are unique for their exclusive and effective usage of male insects as pollinators. The key innovation in this system has obviously been the mimicry of mating signals for the attraction of males, enabling the plants to use an array of new pollination niches, i.e. the pool of male insects available as potential pollinators. Adaptation to different pollinator-males required the diversification of mating signals mimicry according to different mate-communication channels. In this scenario, the obtaining of an actual advantage compared to the ancestral pollination system is not a necessary prerequisite.
Evolutionary consequences: patterns of speciation
Food and sexually deceptive orchids show distinct patterns of speciation and mechanisms of reproductive isolation. Food deceptive species are often less specific in pollinator attraction (Cozzolino et al. in press) and thus postzygotic reproductive isolation may be important for the maintenance of species barriers. In sexually deceptive orchids, however, high specificity in pollination accounts for prezygotic isolation (Paulus and Gack 1990; Bower 1996) and post-mating barriers are often lacking (Ehrendorfer 1980). Accordingly, karyotype differences have been found to be larger in food deceptive than in sexually deceptive orchids (Cozzolino et al. 2004).
Sexually deceptive orchids may represent a prime example of sympatric speciation, because high pollinator specificity is linked to differences in the floral odor. Sympatric speciation is predicted when alleles connected to reproductive isolation, e.g. mutations altering floral odor, are strongly selected and therefore become established in the presence of gene flow (Maynard Smith 2002; Rieseberg et al. 2003). Odor differences among orchid species are often small, e.g. in the closely related Ophrys fusca and O. bilunulata, where a similar set of cuticular hydrocarbons are active in the pollinators, but the relative amounts of the compounds differ significantly from each other (Schiestl and Ayasse 2002). The active floral odor of O. sphegodes also differs from O. exaltata only by the relative abundances of the active compounds (Mant et al. in press c). These data indicate that among certain Ophrys species, a mechanism of pattern recognition of cuticular hydrocarbons in male bees mediates the specificity of pollinator attraction. Changes in the expression of genes responsible for the production of these compounds may thus lead to a change in hydrocarbon pattern and alter pollinator attraction. In Chiloglottis, several species differ only in functional groups of one compound that is responsible for pollinator attraction (Mant and Schiestl unpublished). In this system it seems even more obvious that few mutations can change the floral trait mediating reproductive isolation.
Orchids provide exceptional model organisms for plant-pollinator research with the diversity of such interactions unmatched by any other group of plants, enabling unlimited possibilities for comparative research. Further, male reproductive success can be monitored quantitatively in orchids through tracking of pollen flow, export, and loss using stained pollinia or microtags (Peakall 1989; Nilsson et al. 1992; Johnson et al. 2004). Generally, we know little about how selection varies through female and male function in hermaphroditic plants.
Our understanding of the ultimate causes of the evolution of deception in orchids remains very incomplete, and we need more studies addressing the link between deception and outbreeding, offspring quality, lifetime reproductive success, and the usage of pollinator niches. Comparative investigations with other plant groups having pollinia, e.g. the Asclepiadaceae (now included in the Apocynaceae) may shed light on the role of pollinia in the evolution of deception. Evolutionary consequences of pollination by deception may include sympatric speciation in sexually deceptive species with high-pollinator specificity. On the other hand, gene flow among species may converge sympatric taxa when pollinator-imposed selection varies. Questions relating to strengths and dynamics of selection imposed by pollinators, variation in floral traits and their heritability, extent of gene flow among sympatric species, and hybrid fitness should be addressed to resolve mechanisms of speciation in these fascinating plants. A new and exciting research direction concerns the genetic bases of floral traits linked to reproductive success and-isolation of plants. Although orchids are, through their long generation times and difficulties in breeding less suitable for QTL studies, extended knowledge about gene functions in model species like Arabidopsis and Oryza make it now possible to identify target genes in non-model plants like orchids. Knowledge about genes encoding important floral traits in orchids will give insights into molecular evolution of these genes and reveal historical patterns of selection (Rieseberg et al. 2003).
Orchids are extremely rich in species and speciation rates presumed to be exceptionally high (Gill 1989). Although it has often been hypothesized that this spectacular diversification is linked to the intimate and sometimes bizarre interaction of many species with their pollinators (Darwin 1885), we are still facing the challenge of explaining how these mechanisms work and why they have evolved.
I would like to thank Rod Peakall (Canberra), Jim Mant (Zürich), Salvatore Cozzolino (Naples), Manfred Ayasse (Ulm), Wittko Francke (Hamburg), Fernando Ibarra (Hamburg), Claudia Schulz (Hamburg), and Roman Kaiser (Dübendorf) for sharing their insights during many years of fruitful collaboration. Rod Peakall, Jim Mant, Alex Widmer (Zürich), and Salvatore Cozzolino provided helpful comments on this manuscript, and Charlotte Salzmann (Zürich) supplied a picture for this paper. I would also like to thank Peter Linder (Zürich), Amots Dafni (Haifa), Steve Johnson (Pietermaritzburg), and Hannes Paulus (Vienna), for inspiring discussions on the mysteries of orchid pollination. The research of the author is supported by the SNF (3100–68173.02) and ETH Zürich (02297/41-2710)